Re: [PATCH v11 7/7] KVM: x86: virtualize cpuid faulting

From: Kyle Huey
Date: Wed Nov 16 2016 - 20:59:39 EST


On Wed, Nov 16, 2016 at 12:42 PM, Paolo Bonzini <pbonzini@xxxxxxxxxx> wrote:
>> On Fri, Nov 11, 2016 at 12:54 PM, Nadav Amit <nadav.amit@xxxxxxxxx> wrote:
>> >
>> >> On Nov 10, 2016, at 3:40 PM, Kyle Huey <me@xxxxxxxxxxxx> wrote:
>> >>
>> >> Hardware support for faulting on the cpuid instruction is not required to
>> >> emulate it, because cpuid triggers a VM exit anyways. KVM handles the
>> >> relevant
>> >> MSRs (MSR_PLATFORM_INFO and MSR_MISC_FEATURES_ENABLE) and upon a
>> >> cpuid-induced VM exit checks the cpuid faulting state and the CPL.
>> >> kvm_require_cpl is even kind enough to inject the GP fault for us.
>> >
>> > I think that for consistency you should also revise em_cpuid() of KVM
>> > instruction emulator. Otherwise it may be a potential security
>> > hazard (as far fetched as it currently seems).
>>
>> Hmm, ok. Do you know how I can test this code path?
>
> Try this patch (untested!)

Heh, yes, that would do it. Thanks!

> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index 5382b82462fc..06d940a7c9fa 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -5868,8 +5868,7 @@ static void vmx_set_dr7(struct kvm_vcpu *vcpu, unsigned long val)
>
> static int handle_cpuid(struct kvm_vcpu *vcpu)
> {
> - kvm_emulate_cpuid(vcpu);
> - return 1;
> + return emulate_instruction(vcpu, 0) == EMULATE_DONE;
> }
>
> static int handle_rdmsr(struct kvm_vcpu *vcpu)
>
> Paolo

- Kyle